Adaptive Critic Design for Energy Minimization of Portable Video Communication Devices
|
|
- Bernadette Richardson
- 6 years ago
- Views:
Transcription
1 daptve rtc Desg for Eergy Mmzato of Portable Vdeo ommucato Devces Zhao Su Natoal Isttute of erospace North arola & State Uversty N 74 US X he ad Zhha He Departmet of Electrcal ad omputer Egeerg Uversty of Mssour olumba MO bstract Portable vdeo commucato devces operate o batteres wth lmted eergy supply. However vdeo compresso s computatoally tesve ad eergy-demadg. herefore oe of the cetral challegg ssues portable vdeo commucato system desg s to mmze the eergy cosumpto of vdeo ecodg so as to prolog the operatoal lfetme of portable vdeo devces. I ths work we cosder a vdeo ecoder as a olear system wth a umber of ecoder parameters to ts power cosumpto. We explore the approach of adaptve crtc desg to cotrol ad optmze the power cosumpto behavor of a portable vdeo ecodg system. Our expermetal results demostrate that ths approach s very effcetly beg able to acheve the optmum performace accurately ad robustly. Keywords- vdeo compresso eergy mmzato adaptve cotrol optmzato. I. INRODUION ORBE devces are powered by batteres. Vdeo Pecodg schemes are stll computatoally tesve ad eergy-demadg eve after beg fully optmzed wth exstg software ad hardware eergy-mmzato techques [ 4 8]. s a result the operatoal lfetme of curret portable vdeo systems such as hadheld vdeo devces s stll very short mostly the rage of a few hours. hs has become a bottleeck ssue for techologcal progress portable vdeo electrocs. I the age of desktop computg ad wred commucato people worred about bts storage space or trasmsso badwdth. o aalyze model cotrol ad optmze the performace of a sgal processg ad commucato system uder bt rate costrats ratedstorto (R-D theores ad algorthms have bee developed []. Wth recet techologcal advaces crcut desg ad wreless commucato the storage space ad etwork badwdth have expereced dramatc growth beg mproved by hudreds of tmes durg the past decade. urretly may portable commucato applcatos eergy has become a much more scarce ad crtcal resource tha bts [8]. herefore how to corporate the eergy cosumpto to the exstg R-D performace aalyss framework so as to optmze the commucato system performace uder bt ad eergy costrats emerges as a ew research ssue. here are two types of portable vdeo devces: ecoder (e.g. vdeo cell phoes wreless vdeo cameras etc ad player (e.g. Pod vdeo. I ths work we focus o eergy mmzato for portable vdeo ecodg devces. hs s because o portable vdeo devces the fracto of eergy cosumpto by vdeo ecodg (typcally 6-85% s much hgher tha that of vdeo decodg. o reduce the eergy cosumpto of vdeo ecoders a lot of algorthms software ad hardware eergy-mmzato techques cludg lowcomplexty ecoder desg low-power embedded vdeo ecodg adaptve power cotrol ad jot ecoder ad hardware adaptato have bee developed [ 4 8 9]. hese algorthms focus o ecoder complexty (ad power cosumpto reducto through heurstc adaptato or cotrol stead of systematc eergy optmzato. I ths work we propose to develop a systematc approach to cotrol ad optmze the eergy cosumpto behavor of portable vdeo ecodg devces usg adaptve crtc desg [6]. More specfcally we cosder the vdeo ecoder as a olear dyamc system. he purpose of eergy cosumpto cotrol s to fd a sequece of ecodg parameters such that the overall eergy cosumpto s mmzed uder the rate-dstorto costrats. Our expermetal results demostrate that ths olear system cotrol approach s very effectve beg able to acheve the optmum performace. he rest of the paper s orgazed as follows. I Secto II we dscuss eergy-scalable vdeo ecodg system desg. Secto III presets our adaptve crtc system desg. Secto IV presets the expermetal results o eergy cosumpto cotrol ad optmzato real-tme vdeo ecodg. Secto V cocludes the paper. II. OPERION POWER-RE-DISORION NYSIS I ths secto we study the eergy cosumpto behavor of a vdeo ecoder ad troduce the operatoal power-ratedstorto (P-R-D aalyss. he operatoal P-R-D aalyss wll be performed offle ad provde the groud-truth optmum performace whch wll be used for performace evaluato of the proposed algorthm. he cetral task of P-R-D aalyss s to aswer the followg questo []: what s the mmum vdeo dstorto (or equvaletly vdeo qualty that a vdeo ecoder ca /8/$5. 8 IEEE
2 acheve uder bt rate ad power cosumpto costrats? Fgure : optmum ecoder cotrol parameters. Fgure. P-R-D fucto. Our cetral dea s to troduce a set of complexty cotrol parameters to cotrol (scale dow the computatoal complexty of major ecodg operatos of the vdeo ecoder. Wth DVS (Dyamc Voltage Scalg a recetly developed power cotrol techology for mcroprocessors [] ths complexty-scalable vdeo ecoder ca be traslated to a eergy-scalable vdeo ecoder. More specfcally the eergyscalable vdeo ecoder desg has the followg three major steps. I the frst step we group the ecodg operatos to several modules such as moto predcto pre-codg (trasform ad quatzato ad etropy codg ad the troduce a set of cotrol parameters Γ = [... ] to cotrol the power cosumpto of these modules. herefore the ecoder complexty s the a fucto of these cotrol parameters deoted by (.... Wth the DVS desg framework the ecodg power cosumpto deoted by P s a fucto of ecoder complexty therefore also a fucto of Γ = [... ] deoted by P(.... he expresso of ths fucto also depeds o the power cosumpto model of the specfc mcro-processor. I the thrd step ether expermetally or aalytcally we study the R-D behavor of each cotrol parameter ad tegrate these models to a comprehesve parametrc R-D model for the vdeo ecoder deoted by D( R;.... We perform optmum cofgurato of the cotrol parameters to maxmze the vdeo qualty (or mmze the vdeo dstorto uder the power costrat. hs optmzato problem ca be mathematcally formulated as follows: m D = D( R;... {... } s.t. P... P ( ( where P s the avalable power cosumpto for vdeo ecodg. he optmum soluto deoted by D ( R; P descrbes the P-R-D behavor of the vdeo ecoder. o vew the P-R-D model more detal we plot the D-P curves for dfferet bt rates ragg from. bpp (bts per pxel to. bpp Fg.. Fg. shows the D-P curves at dfferet bt rates R s. Fg. shows the optmum complexty cotrol parameters. We ca see that whe the power supply level s low the D(R fucto s almost flat whch meas the vdeo processg ad ecodg effcecy s very low; hece ths case more badwdth does ot mprove the vdeo presetato qualty. he P-R-D model has drect applcatos eergy maagemet resource allocato ad QoS provsog wreless vdeo commucato especally over wreless vdeo sesor etworks. Fgure : P-R-D fucto.
3 III. DPIVE RII DESIGN FOR ENERGY ONRO ND OPIMIZION It should be oted that the operatoal P-R-D aalyss proposed the prevous secto volves very hgh computatoal complexty. It s teded for offle modelg ad aalyss oly ad ot sutable for real-tme eergy cosumpto cotrol ad optmzato. I ths work based o the acto-depedet ad globalzed dual heurstc dyamc programmg (DGDHP method we desged a optmal cotroller for real-tme eergy cosumpto cotrol ad optmzato [6 7]. he adaptato DGDHP s show Fg. 4. It cossts of three major compoets: a model etwork a crtc etwork ad a acto etwork. he puts to the model etwork are system state varable X ( k = [ D( k P( k R] ad ecoder cotrol parameters k ( = [ ( k ( k ( k ]. Here D(k ad P(k are vdeo codg dstorto ad ecodg power cosumpto at tme (or frame k. We assume a rate cotrol algorthm operates sde the ecoder such that the ecodg bt rate R remas relatvely costat throughout the process. I ths work we set k ( = [ ( k ( k ( k ] to be the complexty cotrol parameters Γ = [... ]. he output of the model etwork s the system state at the ext tme stace. he task of the crtc etwork s to lear the cost-to-go performace metrc fucto J(k whle the acto etwork wll determe the ecoder cotrol parameters k ( = [ ( k ( k ( k ] to mmze J(k. I the followg we wll dscuss detaled desg of these three etworks. Fgure 4: block dagram of the proposed DGDHP desg for eergy cosumpto cotrol ad optmzato.. rtc Network Desg Dyamc programmg s a geeral approach for sequetal optmzato applcable uder very broad codtos. Fudametal to ths approach s Bellma Prcple of Optmalty [5]: f a trajectory of actos s optmum o matter how a termedate pot s reached the rest of the trajectory must cocde wth a optmal trajectory as calculated wth the termedate pot as the startg pot. hs prcple s appled by formulatg a "prmary" utlty fucto U(t that represets a cotrol objectve for a partcular cotext oe or more measurable varables. secodary utlty fucto s the formed: K J ( t = U( t+ k ( k= Ufortuately mmzato of J(t s ot computatoally tractable for most real applcatos. Istead the Bellma recurso s ofte used: J ( t = U ( t + J ( t + ( where s a dscout factor for fte horzo problems ( < r < ad U(t s the utlty fucto or local cost. crtc eural etwork show Fg.5 s used to estmate J(t. Here we chose hdde euro thus = wth weghts w = [ w w w w w w w ]( = V = V V V V s the weght of the output euro. I ths eural etwork we have: q = [ D P R ] W S = σ ( q = (4 q + e J = [ S S S] V he crtc NN outputs the scalar J ad two vectors λ X ( k + ad λ ( k +. J s the cost fucto estmated as: hc J( k = Vc( k S ( k where S ( k = σ ( q( k = q = + e 6 q ( k = wj ( k I j= ( k ( I ( k = ( k j Jk ( + λx = [ λx λ ] [ ] ( X λ X = λ D λp λr whrere λx k + = X ( k + λ = [ λ λ λ ] where λ ( k + = ( k + (5 It should be oted that acto depedet forms of D s we have X( k + = λx ( k + l( k = l( k (6 m X ( k + * l( k = λx ( k + + ( λ k l X j( k = X j( k l= X j( k where * J ( k + U( k λ l ( k = + (7 l( k l( k m are the dmeso of the output of the model ad the acto etworks respectvely. he crtc etwork tres to mmze the followg error measure over tme: ( ( ( E = E ( ( k + ER k ER k + E k E k where k k k
4 E = E ( k + E ( k E ( k + E ( k E ( k x x k k k E ( k = Jk ( Jk ( + Uk ( J( k U( k E ( k = ( E ( k R x x j X j( k X j( k X j( k J( k J( k + U( k E ( k = ( E ( k R m l l( k l( k l( k s the dscout factor ( < <. We apply the MS algorthm ad t results a update rule as below: m J( k J( k J( k + U( k J( k J( k J( k + U( k J( k η Δ w = η E ( k η [ ] + η [ ] w X ( k X ( k X ( k X w ( k ( k ( k w j s t = = s s s s t t t t j j j η η are three postve learg rates wth respect to the three tug paths Fg.. Fgure 5: rtc eural etwork. B. cto Network Desg he acto NN has a smlar structure as that of the crtc NN. Fg. 5 also shows the drect adaptato path λ ( k + betwee the acto ad the crtc etworks. he goal of acto s trag s to make λ (t the dervatve of J(t wth respect to the acto to approach zero. o tra the acto etwork we use oly the crtc s J / outputs so as to meet the equato above. hus the error acto s: E = λ ( k + λ( k + (5 he put of acto NN s the state vector X ad the output s the cotrol or acto vector. he weght update rule ca me smply expressed as follow: k ( Δ W = η [ λ( k + ] (6 W where η s a postve learg rate of the acto NN.. Model Neural Network Desg he objectve of model NN s to predct the ext system state X(k+ for gve curret system state X(k ad acto vector (or ecoder cotrol parameters (k. o do ths we also use the eural etwork show Fg. 5. I vdeo ecodg we are able to drectly measure the system state varables. et Y(k+ = [D(k+ P(k+ R(k+] (7 be the measured system state varables amely the ecoded vdeo qualty D(k+ ecodg bt rate R(k+ ad ecodg power P(k+. he the error modelg s gve by E = [ Y ( k + X ( k + ] [ Y ( k + X ( k + ] (8 whch ca be used to tue the weghts of the model eural etwork. IV. ENERGY ONSUMPION ONRO ND OPIMIZION FOR RE-IME VIDEO ENODING I vdeo ecodg there are three computatoally tesve operatos amely moto estmato pre-codg ad etropy codg. ll of these operatos are performed o a block bass. I ths work we use three ecoder complexty cotrol parameters k ( = [ ( k ( k ( k ] where ( k s the umber of SD (sum of absolute dfferece computatos used moto search; ( k s the umber of skpped blocks; ( k s the frame rate. Here k represets the frame umber. he model s traed offle wth trag sequeces. I ths work we use 6 trag vdeo sequeces all QIF (76x44 sze ecoded by MPEG-4 vdeo ecoder at 5 frames per secod. he proposed DGDHP approach for eergy cosumpto cotrol operates at vdeo frame level. he objectve s to mmze the overall vdeo dstorto uder the eergy costrat. We choose two test QIF vdeo sequeces arphoe ad oastgurad show Fg. 6 to evaluate the proposed approach. Each sequece has frames. Fg. 7 shows the average vdeo dstorto (dotted les acheved by the proposed approach o arphoe. It also shows sold les the groud truth optmum obtaed by operatoal P-R-D aalyss. he result for oastguard s show Fg. 8. It ca be see that the performace gap betwee them s very small. hs dcates that the proposed approach for real-tme eergy cosumpto cotrol ad optmzato s very effectve. V. VI. ONUSION Eergy cosumpto cotrol ad optmzato for real-tme vdeo ecodg over portable vdeo commucato devces s a very challegg task sce the ecoder s hghly olear wth very complex eergy cosumpto behavors. Operatoal P-R-D aalyss s too computatoally tesve ad s ot sutable for real-tme applcato. I ths work we have developed a adaptve crtc desg approach for eergy cosumpto cotrol ad optmzato. Our expermetal results demostrate that ths approach s very effcetly beg able to acheve the optmum performace accurately ad robustly. 4
5 Fgure 6: test vdeo clps: arphoe ad oastguard. Fgure 7: Eergy cosumpto cotrol result ad the groudtruth optmum o arphoe vdeo. []. Skora he MPEG-4 vdeo stadard verfcato model IEEE ras. o rcuts ad Systems for Vdeo echology vol. 7 pp. 9 February 997. [] W. P. Burleso P. Ja ad S. Vekatrama Dyamcally Parameterzed rchtecture for Power- ware Vdeo odg: Moto Estmato ad D Proceedgs of the Secod USF Iteratoal Workshop o Dgtal ad omputatoal Vdeo. [4] D. G. Sachs S. dve D.. Joes ross-layer daptve Vdeo odg to Reduce Eergy o Geeral-Purpose Processors Proceedgs of the Iteratoal oferece o Image Processg (IIP Barceloa Spa September. [5] Bellma R.E. Dyamc Programmg Prceto Uversty Press 957. [6] Prokhorov D. & D. Wusch "daptve rtc Desgs" IEEE rasactos o Neural Networks vol. 8 (5 997 pp [7] Prokhorov D. R. Satago & D. Wusch "daptve rtc Desgs: ase Study for Neurocotrol" Neural Networks vol. 8 (9 pp [8] W. P. Burleso P. Ja ad S. Vekatrama Dyamcally Parameterzed rchtecture for Power- ware Vdeo odg: Moto Estmato ad D Proceedgs of the Secod USF Iteratoal Workshop o Dgtal ad omputatoal Vdeo. [9] M. va der Schaar ad Y. dreopoulos Ratedstorto-complexty Modelg for Network ad Recever ware daptato IEEE rasactos o Multmeda vol. 7 o. pp Jue 5. [] Z. He ad S. K. Mtra Ufed Rate-Dstorto alyss Framework for rasform odg IEEE rasactos o rcuts ad System o Vdeo echology vol. pp. -6 December. Fgure 8: Eergy cosumpto cotrol result ad the groudtruth optmum o oastguard vdeo. KNOWEDGEMEN hs work has bee supported part by NSF uder grat DBI-598. REFERENES [] Z. He Y. ag. he I. hmad ad D. Wu Powerrate-dstorto aalyss for wreless vdeo commucato uder eergy costrat IEEE rasactos o rcuts ad System for Vdeo echology May 5. 5
Solving Constrained Flow-Shop Scheduling. Problems with Three Machines
It J Cotemp Math Sceces, Vol 5, 2010, o 19, 921-929 Solvg Costraed Flow-Shop Schedulg Problems wth Three Maches P Pada ad P Rajedra Departmet of Mathematcs, School of Advaced Sceces, VIT Uversty, Vellore-632
More informationEstimation of Stress- Strength Reliability model using finite mixture of exponential distributions
Iteratoal Joural of Computatoal Egeerg Research Vol, 0 Issue, Estmato of Stress- Stregth Relablty model usg fte mxture of expoetal dstrbutos K.Sadhya, T.S.Umamaheswar Departmet of Mathematcs, Lal Bhadur
More informationSource-Channel Prediction in Error Resilient Video Coding
Source-Chael Predcto Error Reslet Vdeo Codg Hua Yag ad Keeth Rose Sgal Compresso Laboratory ECE Departmet Uversty of Calfora, Sata Barbara Outle Itroducto Source-chael predcto Smulato results Coclusos
More informationLecture 16: Backpropogation Algorithm Neural Networks with smooth activation functions
CO-511: Learg Theory prg 2017 Lecturer: Ro Lv Lecture 16: Bacpropogato Algorthm Dsclamer: These otes have ot bee subected to the usual scruty reserved for formal publcatos. They may be dstrbuted outsde
More informationFunctions of Random Variables
Fuctos of Radom Varables Chapter Fve Fuctos of Radom Varables 5. Itroducto A geeral egeerg aalyss model s show Fg. 5.. The model output (respose) cotas the performaces of a system or product, such as weght,
More informationBlock-Based Compact Thermal Modeling of Semiconductor Integrated Circuits
Block-Based Compact hermal Modelg of Semcoductor Itegrated Crcuts Master s hess Defese Caddate: Jg Ba Commttee Members: Dr. Mg-Cheg Cheg Dr. Daqg Hou Dr. Robert Schllg July 27, 2009 Outle Itroducto Backgroud
More informationAnalysis of Lagrange Interpolation Formula
P IJISET - Iteratoal Joural of Iovatve Scece, Egeerg & Techology, Vol. Issue, December 4. www.jset.com ISS 348 7968 Aalyss of Lagrage Iterpolato Formula Vjay Dahya PDepartmet of MathematcsMaharaja Surajmal
More informationPart 4b Asymptotic Results for MRR2 using PRESS. Recall that the PRESS statistic is a special type of cross validation procedure (see Allen (1971))
art 4b Asymptotc Results for MRR usg RESS Recall that the RESS statstc s a specal type of cross valdato procedure (see Alle (97)) partcular to the regresso problem ad volves fdg Y $,, the estmate at the
More informationA tighter lower bound on the circuit size of the hardest Boolean functions
Electroc Colloquum o Computatoal Complexty, Report No. 86 2011) A tghter lower boud o the crcut sze of the hardest Boolea fuctos Masak Yamamoto Abstract I [IPL2005], Fradse ad Mlterse mproved bouds o the
More informationIntroduction to local (nonparametric) density estimation. methods
Itroducto to local (oparametrc) desty estmato methods A slecture by Yu Lu for ECE 66 Sprg 014 1. Itroducto Ths slecture troduces two local desty estmato methods whch are Parze desty estmato ad k-earest
More informationbest estimate (mean) for X uncertainty or error in the measurement (systematic, random or statistical) best
Error Aalyss Preamble Wheever a measuremet s made, the result followg from that measuremet s always subject to ucertaty The ucertaty ca be reduced by makg several measuremets of the same quatty or by mprovg
More informationFeature Selection: Part 2. 1 Greedy Algorithms (continued from the last lecture)
CSE 546: Mache Learg Lecture 6 Feature Selecto: Part 2 Istructor: Sham Kakade Greedy Algorthms (cotued from the last lecture) There are varety of greedy algorthms ad umerous amg covetos for these algorthms.
More informationKernel-based Methods and Support Vector Machines
Kerel-based Methods ad Support Vector Maches Larr Holder CptS 570 Mache Learg School of Electrcal Egeerg ad Computer Scece Washgto State Uverst Refereces Muller et al. A Itroducto to Kerel-Based Learg
More informationAnalyzing Fuzzy System Reliability Using Vague Set Theory
Iteratoal Joural of Appled Scece ad Egeerg 2003., : 82-88 Aalyzg Fuzzy System Relablty sg Vague Set Theory Shy-Mg Che Departmet of Computer Scece ad Iformato Egeerg, Natoal Tawa versty of Scece ad Techology,
More informationMultiple Choice Test. Chapter Adequacy of Models for Regression
Multple Choce Test Chapter 06.0 Adequac of Models for Regresso. For a lear regresso model to be cosdered adequate, the percetage of scaled resduals that eed to be the rage [-,] s greater tha or equal to
More informationD. VQ WITH 1ST-ORDER LOSSLESS CODING
VARIABLE-RATE VQ (AKA VQ WITH ENTROPY CODING) Varable-Rate VQ = Quatzato + Lossless Varable-Legth Bary Codg A rage of optos -- from smple to complex A. Uform scalar quatzato wth varable-legth codg, oe
More informationDimensionality Reduction and Learning
CMSC 35900 (Sprg 009) Large Scale Learg Lecture: 3 Dmesoalty Reducto ad Learg Istructors: Sham Kakade ad Greg Shakharovch L Supervsed Methods ad Dmesoalty Reducto The theme of these two lectures s that
More informationResearch on SVM Prediction Model Based on Chaos Theory
Advaced Scece ad Techology Letters Vol.3 (SoftTech 06, pp.59-63 http://dx.do.org/0.457/astl.06.3.3 Research o SVM Predcto Model Based o Chaos Theory Sog Lagog, Wu Hux, Zhag Zezhog 3, College of Iformato
More informationAssignment 5/MATH 247/Winter Due: Friday, February 19 in class (!) (answers will be posted right after class)
Assgmet 5/MATH 7/Wter 00 Due: Frday, February 9 class (!) (aswers wll be posted rght after class) As usual, there are peces of text, before the questos [], [], themselves. Recall: For the quadratc form
More informationECON 5360 Class Notes GMM
ECON 560 Class Notes GMM Geeralzed Method of Momets (GMM) I beg by outlg the classcal method of momets techque (Fsher, 95) ad the proceed to geeralzed method of momets (Hase, 98).. radtoal Method of Momets
More informationAn Introduction to. Support Vector Machine
A Itroducto to Support Vector Mache Support Vector Mache (SVM) A classfer derved from statstcal learg theory by Vapk, et al. 99 SVM became famous whe, usg mages as put, t gave accuracy comparable to eural-etwork
More informationA Robust Total Least Mean Square Algorithm For Nonlinear Adaptive Filter
A Robust otal east Mea Square Algorthm For Nolear Adaptve Flter Ruxua We School of Electroc ad Iformato Egeerg X'a Jaotog Uversty X'a 70049, P.R. Cha rxwe@chare.com Chogzhao Ha, azhe u School of Electroc
More informationEconometric Methods. Review of Estimation
Ecoometrc Methods Revew of Estmato Estmatg the populato mea Radom samplg Pot ad terval estmators Lear estmators Ubased estmators Lear Ubased Estmators (LUEs) Effcecy (mmum varace) ad Best Lear Ubased Estmators
More informationUnimodality Tests for Global Optimization of Single Variable Functions Using Statistical Methods
Malaysa Umodalty Joural Tests of Mathematcal for Global Optmzato Sceces (): of 05 Sgle - 5 Varable (007) Fuctos Usg Statstcal Methods Umodalty Tests for Global Optmzato of Sgle Varable Fuctos Usg Statstcal
More informationBayes (Naïve or not) Classifiers: Generative Approach
Logstc regresso Bayes (Naïve or ot) Classfers: Geeratve Approach What do we mea by Geeratve approach: Lear p(y), p(x y) ad the apply bayes rule to compute p(y x) for makg predctos Ths s essetally makg
More informationMA/CSSE 473 Day 27. Dynamic programming
MA/CSSE 473 Day 7 Dyamc Programmg Bomal Coeffcets Warshall's algorthm (Optmal BSTs) Studet questos? Dyamc programmg Used for problems wth recursve solutos ad overlappg subproblems Typcally, we save (memoze)
More informationLecture 12: Multilayer perceptrons II
Lecture : Multlayer perceptros II Bayes dscrmats ad MLPs he role of hdde uts A eample Itroducto to Patter Recoto Rcardo Guterrez-Osua Wrht State Uversty Bayes dscrmats ad MLPs ( As we have see throuhout
More informationRademacher Complexity. Examples
Algorthmc Foudatos of Learg Lecture 3 Rademacher Complexty. Examples Lecturer: Patrck Rebesch Verso: October 16th 018 3.1 Itroducto I the last lecture we troduced the oto of Rademacher complexty ad showed
More informationBeam Warming Second-Order Upwind Method
Beam Warmg Secod-Order Upwd Method Petr Valeta Jauary 6, 015 Ths documet s a part of the assessmet work for the subject 1DRP Dfferetal Equatos o Computer lectured o FNSPE CTU Prague. Abstract Ths documet
More informationVARIABLE-RATE VQ (AKA VQ WITH ENTROPY CODING)
VARIABLE-RATE VQ (AKA VQ WITH ENTROPY CODING) Varable-Rate VQ = Quatzato + Lossless Varable-Legth Bary Codg A rage of optos -- from smple to complex a. Uform scalar quatzato wth varable-legth codg, oe
More informationGeneralization of the Dissimilarity Measure of Fuzzy Sets
Iteratoal Mathematcal Forum 2 2007 o. 68 3395-3400 Geeralzato of the Dssmlarty Measure of Fuzzy Sets Faramarz Faghh Boformatcs Laboratory Naobotechology Research Ceter vesa Research Isttute CECR Tehra
More information{ }{ ( )} (, ) = ( ) ( ) ( ) Chapter 14 Exercises in Sampling Theory. Exercise 1 (Simple random sampling): Solution:
Chapter 4 Exercses Samplg Theory Exercse (Smple radom samplg: Let there be two correlated radom varables X ad A sample of sze s draw from a populato by smple radom samplg wthout replacemet The observed
More informationf f... f 1 n n (ii) Median : It is the value of the middle-most observation(s).
CHAPTER STATISTICS Pots to Remember :. Facts or fgures, collected wth a defte pupose, are called Data.. Statstcs s the area of study dealg wth the collecto, presetato, aalyss ad terpretato of data.. The
More informationConsensus Control for a Class of High Order System via Sliding Mode Control
Cosesus Cotrol for a Class of Hgh Order System va Sldg Mode Cotrol Chagb L, Y He, ad Aguo Wu School of Electrcal ad Automato Egeerg, Taj Uversty, Taj, Cha, 300072 Abstract. I ths paper, cosesus problem
More informationRegression and the LMS Algorithm
CSE 556: Itroducto to Neural Netorks Regresso ad the LMS Algorthm CSE 556: Regresso 1 Problem statemet CSE 556: Regresso Lear regresso th oe varable Gve a set of N pars of data {, d }, appromate d b a
More information13. Parametric and Non-Parametric Uncertainties, Radial Basis Functions and Neural Network Approximations
Lecture 7 3. Parametrc ad No-Parametrc Ucertates, Radal Bass Fuctos ad Neural Network Approxmatos he parameter estmato algorthms descrbed prevous sectos were based o the assumpto that the system ucertates
More informationBounds on the expected entropy and KL-divergence of sampled multinomial distributions. Brandon C. Roy
Bouds o the expected etropy ad KL-dvergece of sampled multomal dstrbutos Brado C. Roy bcroy@meda.mt.edu Orgal: May 18, 2011 Revsed: Jue 6, 2011 Abstract Iformato theoretc quattes calculated from a sampled
More informationChapter 9 Jordan Block Matrices
Chapter 9 Jorda Block atrces I ths chapter we wll solve the followg problem. Gve a lear operator T fd a bass R of F such that the matrx R (T) s as smple as possble. f course smple s a matter of taste.
More informationA Method for Damping Estimation Based On Least Square Fit
Amerca Joural of Egeerg Research (AJER) 5 Amerca Joural of Egeerg Research (AJER) e-issn: 3-847 p-issn : 3-936 Volume-4, Issue-7, pp-5-9 www.ajer.org Research Paper Ope Access A Method for Dampg Estmato
More informationA New Family of Transformations for Lifetime Data
Proceedgs of the World Cogress o Egeerg 4 Vol I, WCE 4, July - 4, 4, Lodo, U.K. A New Famly of Trasformatos for Lfetme Data Lakhaa Watthaacheewakul Abstract A famly of trasformatos s the oe of several
More informationLecture 12 APPROXIMATION OF FIRST ORDER DERIVATIVES
FDM: Appromato of Frst Order Dervatves Lecture APPROXIMATION OF FIRST ORDER DERIVATIVES. INTRODUCTION Covectve term coservato equatos volve frst order dervatves. The smplest possble approach for dscretzato
More informationThe Mathematical Appendix
The Mathematcal Appedx Defto A: If ( Λ, Ω, where ( λ λ λ whch the probablty dstrbutos,,..., Defto A. uppose that ( Λ,,..., s a expermet type, the σ-algebra o λ λ λ are defed s deoted by ( (,,...,, σ Ω.
More informationCSE 5526: Introduction to Neural Networks Linear Regression
CSE 556: Itroducto to Neural Netorks Lear Regresso Part II 1 Problem statemet Part II Problem statemet Part II 3 Lear regresso th oe varable Gve a set of N pars of data , appromate d by a lear fucto
More informationQuantization in Dynamic Smarandache Multi-Space
Quatzato Dyamc Smaradache Mult-Space Fu Yuhua Cha Offshore Ol Research Ceter, Beg, 7, Cha (E-mal: fuyh@cooc.com.c ) Abstract: Dscussg the applcatos of Dyamc Smaradache Mult-Space (DSMS) Theory. Supposg
More informationAnalysis of System Performance IN2072 Chapter 5 Analysis of Non Markov Systems
Char for Network Archtectures ad Servces Prof. Carle Departmet of Computer Scece U Müche Aalyss of System Performace IN2072 Chapter 5 Aalyss of No Markov Systems Dr. Alexader Kle Prof. Dr.-Ig. Georg Carle
More information2.160 System Identification, Estimation, and Learning Lecture Notes No. 17 April 24, 2006
.6 System Idetfcato, Estmato, ad Learg Lectre Notes No. 7 Aprl 4, 6. Iformatve Expermets. Persstece of Exctato Iformatve data sets are closely related to Persstece of Exctato, a mportat cocept sed adaptve
More information1 Mixed Quantum State. 2 Density Matrix. CS Density Matrices, von Neumann Entropy 3/7/07 Spring 2007 Lecture 13. ψ = α x x. ρ = p i ψ i ψ i.
CS 94- Desty Matrces, vo Neuma Etropy 3/7/07 Sprg 007 Lecture 3 I ths lecture, we wll dscuss the bascs of quatum formato theory I partcular, we wll dscuss mxed quatum states, desty matrces, vo Neuma etropy
More informationResearch on Fault Tolerance for the Static Segment of FlexRay Protocol
Research o Fault Tolerace for the Statc Segmet of FlexRay Protocol Ru I a, Ye ZHU a, Zhyg WANG b a Embedded System & Networkg aboratory, Hua Uversty, Cha b the School of Computer, Natoal Uversty of Defese
More informationBERNSTEIN COLLOCATION METHOD FOR SOLVING NONLINEAR DIFFERENTIAL EQUATIONS. Aysegul Akyuz Dascioglu and Nese Isler
Mathematcal ad Computatoal Applcatos, Vol. 8, No. 3, pp. 293-300, 203 BERNSTEIN COLLOCATION METHOD FOR SOLVING NONLINEAR DIFFERENTIAL EQUATIONS Aysegul Ayuz Dascoglu ad Nese Isler Departmet of Mathematcs,
More informationhp calculators HP 30S Statistics Averages and Standard Deviations Average and Standard Deviation Practice Finding Averages and Standard Deviations
HP 30S Statstcs Averages ad Stadard Devatos Average ad Stadard Devato Practce Fdg Averages ad Stadard Devatos HP 30S Statstcs Averages ad Stadard Devatos Average ad stadard devato The HP 30S provdes several
More information13. Artificial Neural Networks for Function Approximation
Lecture 7 3. Artfcal eural etworks for Fucto Approxmato Motvato. A typcal cotrol desg process starts wth modelg, whch s bascally the process of costructg a mathematcal descrpto (such as a set of ODE-s)
More informationAn Indian Journal FULL PAPER ABSTRACT KEYWORDS. Trade Science Inc. Research on scheme evaluation method of automation mechatronic systems
[ype text] [ype text] [ype text] ISSN : 0974-7435 Volume 0 Issue 6 Boechology 204 Ida Joural FULL PPER BIJ, 0(6, 204 [927-9275] Research o scheme evaluato method of automato mechatroc systems BSRC Che
More information1. A real number x is represented approximately by , and we are told that the relative error is 0.1 %. What is x? Note: There are two answers.
PROBLEMS A real umber s represeted appromately by 63, ad we are told that the relatve error s % What s? Note: There are two aswers Ht : Recall that % relatve error s What s the relatve error volved roudg
More informationStudy on a Fire Detection System Based on Support Vector Machine
Sesors & Trasducers, Vol. 8, Issue, November 04, pp. 57-6 Sesors & Trasducers 04 by IFSA Publshg, S. L. http://www.sesorsportal.com Study o a Fre Detecto System Based o Support Vector Mache Ye Xaotg, Wu
More informationA conic cutting surface method for linear-quadraticsemidefinite
A coc cuttg surface method for lear-quadratcsemdefte programmg Mohammad R. Osoorouch Calfora State Uversty Sa Marcos Sa Marcos, CA Jot wor wth Joh E. Mtchell RPI July 3, 2008 Outle: Secod-order coe: defto
More informationX X X E[ ] E X E X. is the ()m n where the ( i,)th. j element is the mean of the ( i,)th., then
Secto 5 Vectors of Radom Varables Whe workg wth several radom varables,,..., to arrage them vector form x, t s ofte coveet We ca the make use of matrx algebra to help us orgaze ad mapulate large umbers
More informationABOUT ONE APPROACH TO APPROXIMATION OF CONTINUOUS FUNCTION BY THREE-LAYERED NEURAL NETWORK
ABOUT ONE APPROACH TO APPROXIMATION OF CONTINUOUS FUNCTION BY THREE-LAYERED NEURAL NETWORK Ram Rzayev Cyberetc Isttute of the Natoal Scece Academy of Azerbaa Republc ramrza@yahoo.com Aygu Alasgarova Khazar
More informationUnsupervised Learning and Other Neural Networks
CSE 53 Soft Computg NOT PART OF THE FINAL Usupervsed Learg ad Other Neural Networs Itroducto Mture Destes ad Idetfablty ML Estmates Applcato to Normal Mtures Other Neural Networs Itroducto Prevously, all
More information18.413: Error Correcting Codes Lab March 2, Lecture 8
18.413: Error Correctg Codes Lab March 2, 2004 Lecturer: Dael A. Spelma Lecture 8 8.1 Vector Spaces A set C {0, 1} s a vector space f for x all C ad y C, x + y C, where we take addto to be compoet wse
More informationLecture 7. Confidence Intervals and Hypothesis Tests in the Simple CLR Model
Lecture 7. Cofdece Itervals ad Hypothess Tests the Smple CLR Model I lecture 6 we troduced the Classcal Lear Regresso (CLR) model that s the radom expermet of whch the data Y,,, K, are the outcomes. The
More informationChapter 5 Properties of a Random Sample
Lecture 6 o BST 63: Statstcal Theory I Ku Zhag, /0/008 Revew for the prevous lecture Cocepts: t-dstrbuto, F-dstrbuto Theorems: Dstrbutos of sample mea ad sample varace, relatoshp betwee sample mea ad sample
More informationCan we take the Mysticism Out of the Pearson Coefficient of Linear Correlation?
Ca we tae the Mstcsm Out of the Pearso Coeffcet of Lear Correlato? Itroducto As the ttle of ths tutoral dcates, our purpose s to egeder a clear uderstadg of the Pearso coeffcet of lear correlato studets
More informationNon-uniform Turán-type problems
Joural of Combatoral Theory, Seres A 111 2005 106 110 wwwelsevercomlocatecta No-uform Turá-type problems DhruvMubay 1, Y Zhao 2 Departmet of Mathematcs, Statstcs, ad Computer Scece, Uversty of Illos at
More informationECE 595, Section 10 Numerical Simulations Lecture 19: FEM for Electronic Transport. Prof. Peter Bermel February 22, 2013
ECE 595, Secto 0 Numercal Smulatos Lecture 9: FEM for Electroc Trasport Prof. Peter Bermel February, 03 Outle Recap from Wedesday Physcs-based devce modelg Electroc trasport theory FEM electroc trasport
More informationSTK4011 and STK9011 Autumn 2016
STK4 ad STK9 Autum 6 Pot estmato Covers (most of the followg materal from chapter 7: Secto 7.: pages 3-3 Secto 7..: pages 3-33 Secto 7..: pages 35-3 Secto 7..3: pages 34-35 Secto 7.3.: pages 33-33 Secto
More informationLecture 3. Sampling, sampling distributions, and parameter estimation
Lecture 3 Samplg, samplg dstrbutos, ad parameter estmato Samplg Defto Populato s defed as the collecto of all the possble observatos of terest. The collecto of observatos we take from the populato s called
More informationRuntime analysis RLS on OneMax. Heuristic Optimization
Lecture 6 Rutme aalyss RLS o OeMax trals of {,, },, l ( + ɛ) l ( ɛ)( ) l Algorthm Egeerg Group Hasso Platter Isttute, Uversty of Potsdam 9 May T, We wat to rgorously uderstad ths behavor 9 May / Rutme
More informationLECTURE - 4 SIMPLE RANDOM SAMPLING DR. SHALABH DEPARTMENT OF MATHEMATICS AND STATISTICS INDIAN INSTITUTE OF TECHNOLOGY KANPUR
amplg Theory MODULE II LECTURE - 4 IMPLE RADOM AMPLIG DR. HALABH DEPARTMET OF MATHEMATIC AD TATITIC IDIA ITITUTE OF TECHOLOGY KAPUR Estmato of populato mea ad populato varace Oe of the ma objectves after
More informationMedian as a Weighted Arithmetic Mean of All Sample Observations
Meda as a Weghted Arthmetc Mea of All Sample Observatos SK Mshra Dept. of Ecoomcs NEHU, Shllog (Ida). Itroducto: Iumerably may textbooks Statstcs explctly meto that oe of the weakesses (or propertes) of
More information1 Lyapunov Stability Theory
Lyapuov Stablty heory I ths secto we cosder proofs of stablty of equlbra of autoomous systems. hs s stadard theory for olear systems, ad oe of the most mportat tools the aalyss of olear systems. It may
More informationOPTIMAL LAY-OUT OF NATURAL GAS PIPELINE NETWORK
23rd World Gas Coferece, Amsterdam 2006 OPTIMAL LAY-OUT OF NATURAL GAS PIPELINE NETWORK Ma author Tg-zhe, Ne CHINA ABSTRACT I cha, there are lots of gas ppele etwork eeded to be desged ad costructed owadays.
More informationNonlinear Blind Source Separation Using Hybrid Neural Networks*
Nolear Bld Source Separato Usg Hybrd Neural Networks* Chu-Hou Zheg,2, Zh-Ka Huag,2, chael R. Lyu 3, ad Tat-g Lok 4 Itellget Computg Lab, Isttute of Itellget aches, Chese Academy of Sceces, P.O.Box 3, Hefe,
More informationAnalysis of Variance with Weibull Data
Aalyss of Varace wth Webull Data Lahaa Watthaacheewaul Abstract I statstcal data aalyss by aalyss of varace, the usual basc assumptos are that the model s addtve ad the errors are radomly, depedetly, ad
More informationCubic Nonpolynomial Spline Approach to the Solution of a Second Order Two-Point Boundary Value Problem
Joural of Amerca Scece ;6( Cubc Nopolyomal Sple Approach to the Soluto of a Secod Order Two-Pot Boudary Value Problem W.K. Zahra, F.A. Abd El-Salam, A.A. El-Sabbagh ad Z.A. ZAk * Departmet of Egeerg athematcs
More informationWireless Link Properties
Opportustc Ecrypto for Robust Wreless Securty R. Chadramoul ( Moul ) moul@steves.edu Multmeda System, Networkg, ad Commucatos (MSyNC) Laboratory, Departmet of Electrcal ad Computer Egeerg, Steves Isttute
More informationInvestigation of Partially Conditional RP Model with Response Error. Ed Stanek
Partally Codtoal Radom Permutato Model 7- vestgato of Partally Codtoal RP Model wth Respose Error TRODUCTO Ed Staek We explore the predctor that wll result a smple radom sample wth respose error whe a
More informationMULTIDIMENSIONAL HETEROGENEOUS VARIABLE PREDICTION BASED ON EXPERTS STATEMENTS. Gennadiy Lbov, Maxim Gerasimov
Iteratoal Boo Seres "Iformato Scece ad Computg" 97 MULTIIMNSIONAL HTROGNOUS VARIABL PRICTION BAS ON PRTS STATMNTS Geady Lbov Maxm Gerasmov Abstract: I the wors [ ] we proposed a approach of formg a cosesus
More informationG S Power Flow Solution
G S Power Flow Soluto P Q I y y * 0 1, Y y Y 0 y Y Y 1, P Q ( k) ( k) * ( k 1) 1, Y Y PQ buses * 1 P Q Y ( k1) *( k) ( k) Q Im[ Y ] 1 P buses & Slack bus ( k 1) *( k) ( k) Y 1 P Re[ ] Slack bus 17 Calculato
More informationSome Different Perspectives on Linear Least Squares
Soe Dfferet Perspectves o Lear Least Squares A stadard proble statstcs s to easure a respose or depedet varable, y, at fed values of oe or ore depedet varables. Soetes there ests a deterstc odel y f (,,
More informationCHAPTER VI Statistical Analysis of Experimental Data
Chapter VI Statstcal Aalyss of Expermetal Data CHAPTER VI Statstcal Aalyss of Expermetal Data Measuremets do ot lead to a uque value. Ths s a result of the multtude of errors (maly radom errors) that ca
More informationECE 559: Wireless Communication Project Report Diversity Multiplexing Tradeoff in MIMO Channels with partial CSIT. Hoa Pham
ECE 559: Wreless Commucato Project Report Dversty Multplexg Tradeoff MIMO Chaels wth partal CSIT Hoa Pham. Summary I ths project, I have studed the performace ga of MIMO systems. There are two types of
More informationA Helmholtz energy equation of state for calculating the thermodynamic properties of fluid mixtures
A Helmholtz eergy equato of state for calculatg the thermodyamc propertes of flud mxtures Erc W. Lemmo, Reer Tller-Roth Abstract New Approach based o hghly accurate EOS for the pure compoets combed at
More informationarxiv: v1 [cs.lg] 22 Feb 2015
SDCA wthout Dualty Sha Shalev-Shwartz arxv:50.0677v cs.lg Feb 05 Abstract Stochastc Dual Coordate Ascet s a popular method for solvg regularzed loss mmzato for the case of covex losses. I ths paper we
More informationDerivation of 3-Point Block Method Formula for Solving First Order Stiff Ordinary Differential Equations
Dervato of -Pot Block Method Formula for Solvg Frst Order Stff Ordary Dfferetal Equatos Kharul Hamd Kharul Auar, Kharl Iskadar Othma, Zara Bb Ibrahm Abstract Dervato of pot block method formula wth costat
More informationSystematic Selection of Parameters in the development of Feedforward Artificial Neural Network Models through Conventional and Intelligent Algorithms
THALES Project No. 65/3 Systematc Selecto of Parameters the developmet of Feedforward Artfcal Neural Network Models through Covetoal ad Itellget Algorthms Research Team G.-C. Vosakos, T. Gaakaks, A. Krmpes,
More informationBasics of Information Theory: Markku Juntti. Basic concepts and tools 1 Introduction 2 Entropy, relative entropy and mutual information
: Maru Jutt Overvew he propertes of adlmted Gaussa chaels are further studed, parallel Gaussa chaels ad Gaussa chaels wth feedac are solved. Source he materal s maly ased o Sectos.4.6 of the course oo
More information1 Onto functions and bijections Applications to Counting
1 Oto fuctos ad bectos Applcatos to Coutg Now we move o to a ew topc. Defto 1.1 (Surecto. A fucto f : A B s sad to be surectve or oto f for each b B there s some a A so that f(a B. What are examples of
More informationClass 13,14 June 17, 19, 2015
Class 3,4 Jue 7, 9, 05 Pla for Class3,4:. Samplg dstrbuto of sample mea. The Cetral Lmt Theorem (CLT). Cofdece terval for ukow mea.. Samplg Dstrbuto for Sample mea. Methods used are based o CLT ( Cetral
More informationarxiv:math/ v1 [math.gm] 8 Dec 2005
arxv:math/05272v [math.gm] 8 Dec 2005 A GENERALIZATION OF AN INEQUALITY FROM IMO 2005 NIKOLAI NIKOLOV The preset paper was spred by the thrd problem from the IMO 2005. A specal award was gve to Yure Boreko
More informationESS Line Fitting
ESS 5 014 17. Le Fttg A very commo problem data aalyss s lookg for relatoshpetwee dfferet parameters ad fttg les or surfaces to data. The smplest example s fttg a straght le ad we wll dscuss that here
More informationLINEARLY CONSTRAINED MINIMIZATION BY USING NEWTON S METHOD
Jural Karya Asl Loreka Ahl Matematk Vol 8 o 205 Page 084-088 Jural Karya Asl Loreka Ahl Matematk LIEARLY COSTRAIED MIIMIZATIO BY USIG EWTO S METHOD Yosza B Dasrl, a Ismal B Moh 2 Faculty Electrocs a Computer
More informationHomework 1: Solutions Sid Banerjee Problem 1: (Practice with Asymptotic Notation) ORIE 4520: Stochastics at Scale Fall 2015
Fall 05 Homework : Solutos Problem : (Practce wth Asymptotc Notato) A essetal requremet for uderstadg scalg behavor s comfort wth asymptotc (or bg-o ) otato. I ths problem, you wll prove some basc facts
More informationDERIVATIVE ESTIMATION WITH KNOWN CONTROL-VARIATE VARIANCES. Jamie R. Wieland Bruce W. Schmeiser
Proceedgs of the 007 Wter Smulato Coferece S G Hederso, B Bller, M-H Hseh, J Shortle, J D Tew, ad R R Barto, eds DERIVATIVE ESTIMATION WITH KNOWN CONTROL-VARIATE VARIANCES Jame R Welad Bruce W Schmeser
More informationL5 Polynomial / Spline Curves
L5 Polyomal / Sple Curves Cotets Coc sectos Polyomal Curves Hermte Curves Bezer Curves B-Sples No-Uform Ratoal B-Sples (NURBS) Mapulato ad Represetato of Curves Types of Curve Equatos Implct: Descrbe a
More informationSPECIAL CONSIDERATIONS FOR VOLUMETRIC Z-TEST FOR PROPORTIONS
SPECIAL CONSIDERAIONS FOR VOLUMERIC Z-ES FOR PROPORIONS Oe s stctve reacto to the questo of whether two percetages are sgfcatly dfferet from each other s to treat them as f they were proportos whch the
More informationResearch Article A New Derivation and Recursive Algorithm Based on Wronskian Matrix for Vandermonde Inverse Matrix
Mathematcal Problems Egeerg Volume 05 Artcle ID 94757 7 pages http://ddoorg/055/05/94757 Research Artcle A New Dervato ad Recursve Algorthm Based o Wroska Matr for Vadermode Iverse Matr Qu Zhou Xja Zhag
More informationDimensionality reduction Feature selection
CS 750 Mache Learg Lecture 3 Dmesoalty reducto Feature selecto Mlos Hauskrecht mlos@cs.ptt.edu 539 Seott Square CS 750 Mache Learg Dmesoalty reducto. Motvato. Classfcato problem eample: We have a put data
More informationChapter 14 Logistic Regression Models
Chapter 4 Logstc Regresso Models I the lear regresso model X β + ε, there are two types of varables explaatory varables X, X,, X k ad study varable y These varables ca be measured o a cotuous scale as
More informationLecture 02: Bounding tail distributions of a random variable
CSCI-B609: A Theorst s Toolkt, Fall 206 Aug 25 Lecture 02: Boudg tal dstrbutos of a radom varable Lecturer: Yua Zhou Scrbe: Yua Xe & Yua Zhou Let us cosder the ubased co flps aga. I.e. let the outcome
More informationApplication of Calibration Approach for Regression Coefficient Estimation under Two-stage Sampling Design
Authors: Pradp Basak, Kaustav Adtya, Hukum Chadra ad U.C. Sud Applcato of Calbrato Approach for Regresso Coeffcet Estmato uder Two-stage Samplg Desg Pradp Basak, Kaustav Adtya, Hukum Chadra ad U.C. Sud
More information